@loop-engine/guards

@loop-engine/guards provides deterministic policy checks that run inside transition execution.

Install

1npm install @loop-engine/guards

Guard model

Guards in this package are synchronous policy assertions wrapped as async functions. They do not call LLMs or network services.

GuardResult:

1interface GuardResult {
2  passed: boolean
3  code?: string
4  message?: string
5  metadata?: Record<string, unknown>
6}

Registry API

1class GuardRegistry {
2  register(guardId: string, evaluator: GuardEvaluator): void
3  get(guardId: string): GuardEvaluator | undefined
4  registerBuiltIns(): void
5}

registerBuiltIns() populates the registry with every guard exported from @loop-engine/guards (see "Built-in guards" below). Call it once on a fresh registry, or use the pre-populated defaultRegistry constant.

1import { createGuardRegistry } from "@loop-engine/guards"
2import { guardId } from "@loop-engine/core"
3 
4const registry = createGuardRegistry()
5registry.register(guardId("budget_available"), {
6  async evaluate(context) {
7    return {
8      passed: context.evidence?.budget_ok === true,
9      message: "Budget check failed"
10    }
11  }
12})

Built-in guards

defaultRegistry (and GuardRegistry.registerBuiltIns()) pre-registers:

• confidence-threshold (ConfidenceThresholdGuard)
• human-only (HumanOnlyGuard)
• evidence-required (EvidenceRequiredGuard)
• cooldown (CooldownGuard)

Hard and soft behavior

Hard/soft severity is defined on each GuardSpec in @loop-engine/core. Runtime blocks on hard failures and records soft failures under _softGuardWarnings in transition evidence.